syi

nkstin,

Received 25 April 2013Received in revised form8 July 2013

Keywords:Spiral cleavageLeech embryoCytokinesis

s byplan through the bilaterally symmetric divisions of mesodermal proteloblast DM and ectodermalproteloblast DNOPQ. We here show that transcriptional inhibitors -amanitin and actinomycin Dspecically disrupt the symmetry and orientation of these two proteloblast cell divisions while having no

y embr

known as spiral99). Spiral clea-trochozoa (Dunnbryo undergoes

ns that alternaterclockwise) chir-tain specic cellndergoing spiraltric cell divisions

This transition to bilateral symmetry is depicted in Fig. 1. At the

Contents lists available at ScienceDirect

journal homepage: www.elsevier.co

Developmen

n Corresponding author. Fax: +1 217 244 1648.

1 Present address: Center for Computational Biology and Bioinformatics,

Developmental Biology 382 (2013) 149159(Holton et al., 1994) and are encircled by a cortical cytoplasm601 S. Goodwin Avenue, Urbana, IL 61801, USA.fourth round of embryonic cleavage, macromere D undergoes aspiral division of sinistral orientation to produce cells DM andDNOPQ (Fig. 1B). These two cells are distinctive in that they bothinherit a pool of yolk-free teloplasm enriched in maternal RNA

0012-1606/$ - see front matter & 2013 Elsevier Inc. All rights reserved.http://dx.doi.org/10.1016/j.ydbio.2013.07.015

University of Texas at Austin, Austin, TX 78712, USA.2 Present address: Department of Molecular and Cell Biology, University of

California, Berkeley, Berkeley, CA 94720, USA.3 Present address: Department of Cell and Developmental Biology, B107

Chemical and Life Sciences Laboratory, University of Illinois at Urbana-Champaign,that pregure the symmetry properties of the larval and adultbody plans (Henry and Martindale, 1999).

E-mail addresses: leechman@austin.utexas.edu (M.W. Schmerer),ryan.null@berkeley.edu (R.W. Null), hastypig@life.illinois.edu (M. Shankland).populations that display coordinated patterns of mitosis (Newportand Kirschner, 1982; OFarrell et al., 1989), whereas other taxadisplay intricately stereotyped patterns of cell division in whicheach individual cytokinesis manifests a specic symmetry andorientation (Sulston et al., 1983). Much has been learned about themolecular pathways that control cleavage geometry (Bergmannet al., 2003; Hampoelz and Knoblich, 2004; Goldstein and Macara,2007), but relatively little is known about the factors that guide a

Leeches are annelid worms, and develop by meanand widely conserved mode of embryogenesiscleavage (reviewed by Henry and Martindale, 19vage is only found in the superphylum Lophoet al., 2008), and is so named because the early emseveral rounds of obliquely oriented cell divisiobetween dextral (clockwise) and sinistral (counteality. But as the spiralian embryo matures, cerlineages in the dorsal or D quadrant stop udivisions and switch instead to bilaterally symmetightly regulated in many species. Some embryos have large cell in the embryonic cell lineage of the leech Helobdella austinensis.s of a distinctiveZygotic transcriptionPax geneDominant-negative

Introduction

The timing and geometry of earlapparent effect on the timing or geometry of other divisions. Transcriptional inhibition had a similareffect on both proteloblasts, i.e. cytokinesis was highly asymmetric and the cleavage plane roughlyorthogonal to that seen during normal development. These ndings suggest that zygotic gene product(s) are required, either directly or indirectly, for the correct placement of the proteloblast cleavagefurrow. The same phenotypes were also observed following in vivo expression of dominant-negative Paxgene constructs. These dominant-negative phenotypes depended on protein/DNA interaction, and couldbe rescued by coexpression of full length Pax proteins. However, symmetric cleavage of the mesodermalproteloblast was rescued by full length constructs of either Hau-Pax1 or Hau-Pax2/5/8, while only Hau-Pax1 rescued the symmetry of ectodermal cleavage. We conclude that both proteloblasts need Pax-mediated transcription to adopt their normally symmetric cleavage patterns, but differ in terms of thespecic Pax proteins required. The implication of these ndings for the evolution of spiral cleavage isdiscussed.

& 2013 Elsevier Inc. All rights reserved.

yonic cell divisions are

given cell lineage through a sequence of different geometries asdevelopment proceeds.

In this paper we investigate one such developmental transitionAccepted 17 July 2013Available online 25 July 2013Developmental transition to bilaterallyregulated by Pax-mediated transcriptionHelobdella austinensis

Matthew W. Schmerer 1, Ryan W. Null 2, Marty ShaSection of Molecular Cell and Developmental Biology, University of Texas at Austin, Au

a r t i c l e i n f o

Article history:

a b s t r a c t

The leech embryo developmmetric cell divisions isn embryos of the leech

land n,3

TX 78712, USA

spiral cleavage, and establishes the symmetry properties of its adult body

m/locate/developmentalbiology

tal Biology

M.W. Schmerer et al. / Developmental Biology 382 (2013) 149159150Fig. 1. Onset of bilaterally symmetric cleavage in the leech Helobdella. An overviewof Helobdella development can be found in Weisblat and Huang (2001): (A) Celllineage leading up to the proteloblast cleavages (bold). Time is hours after rstcleavage for H. austinensis at 24 1C. (B) Animal pole view of embryo at onset of stage4b. Cell D cleaves obliquely to produce DM and DNOPQ, each of which inherits apool of teloplasm (solid gray) encircled by a yolk-rich cortex (stippled). (C) At theonset of stage 5, ectodermal proteloblast DNOPQ cleaves symmetrically toproduce left and right NOPQ cells. Mesodermal proteloblast DM has alreadycleaved symmetrically to produce left and right M teloblasts. The planes ofheavily invested with yolk platelets. Cell DM undergoes a sequenceof two asymmetric, micromere-producing divisions (Fig. 1A), thenits large vegetal granddaughter DM switches to a bilaterallysymmetric division in which the cleavage plane is parallel to theanimal-vegetal (AV) axis (Fig. 1C). This symmetric division segre-gates teloplasm and cortex equally between daughter cells Ml andMr, mesoteloblasts which go on to generate the left and righthalves of the mesoderm (Weisblat and Huang, 2001). We hererefer to DM as the leechs mesodermal proteloblast, and notethat it is functionally equivalent to the 4d cell of standard spiraliannomenclature (Gline et al., 2011).

Cell DNOPQ is situated in the animal hemisphere, and is homo-logous to the 2d cell of other spiralians (Meyer and Seaver, 2011). TheDNOPQ lineage undergoes three highly asymmetric divisions ofalternating chirality to produce yolk-free micromeres dnopq-dnopq (Fig. 1, S1), but its large vegetal great-granddaughter DNOPQ thenswitches to a bilaterally symmetric division whose cleavage plane isclosely parallel to the AV axis (Fig. 1C). This symmetric divisionsegregates teloplasm and cortex equally between daughters NOPQland NOPQr, which will undergo several additional rounds of sym-metric cleavage to produce a bilateral array of ectoteloblasts(Weisblat and Huang, 2001). Cell DNOPQ is therefore said to bean ectodermal proteloblast.

We here report that certain Pax gene transcription factorsplay a critical role in regulating the bilateral symmetry of thesetwo proteloblast cleavages. Pax genes are found throughoutmetazoan animals (Matus et al., 2007), and are known to playimportant roles in development, differentiation, and disease(Lang et al., 2007; Wang et al., 2008). Most Pax proteins arecharacterized by a DNA-binding motif called the paired domain,although some families display a downstream homeodomain as

we microinjected leech embryos with an aqueous solution to

mesodermal and ectodermal symmetry are not yet aligned at this stage. Theindividual cells of the micromere cap (mc) are not delineated in this panel exceptfor dnopqdnopq.which 2% fast green FCF (Sigma) had been added to visuallymonitor injection. Controls were injected with fast green alone.We calculated our standard injection as 0.4% of embryo volume,and varied -amanitin in the injection needle to yield estimatedintraembryonic concentrations of 40 nM or 1 M. Amanitin readilypasses between leech blastomeres (Bissen and Weisblat, 1991),and injection of 1-, 2-, and 4-cell embryos gave indistinguishableresults.

Actinomycin D (Sigma) was dissolved in DMSO as a 40 mg/mlstock, then diluted to 200 g/ml in embryo culture medium.Embryos were treated at the 1- to 4-cell stages, and raised in thedark. Controls were raised in culture medium supplemented with0.5% DMSO.

Lineage tracing and histology

Cleavage-stage embryos were xed overnight with continualagitation in a biphasic mixture of 6.4% formaldehyde (Pella Co.) inwell (Balczarek et al., 1997). There are ve major Pax genefamilies conserved among most bilaterian animals (Matus et al.,2007), but we recently described a sixth family, Pax, thatappears to have arisen by duplication and divergence of thePax2/5/8 gene in an early spiralian ancestor (Schmerer et al.,2009). The Pax gene underwent an additional duplication inthe leechs evolutionary lineage, but only Hau-Pax1 and notits paralog Hau-Pax2 is expressed during the early cleavagestages of embryogenesis in H. austinensis (Schmerer et al., 2009).

In this paper we analyze the developmental function of Hau-Pax1 and two other Pax genes known to have early embryonicexpression. Since Pax proteins function primarily as transcriptionfactors, we also use pharmacological inhibitors to examine thedevelopmental signicance of zygotic transcription per se. Ourresults suggest that the Pax-mediated transcription of one or morezygotic target genes is necessary for the leechs mesodermal andectodermal proteloblasts to make their normal transition tobilateral symmetry.

Methods

Embryos

Leech embryos were taken from breeding colonies maintainedat ambient temperature in 1% articial sea water and fed thriceweekly on pond snails. Embryos were raised at 24 1C in bufferedsaline culture medium, and staged according to Weisblat andHuang (2001). The H. austinensis colony was established bySeaver and Shankland (2000), but Bely and Weisblat (2006) laterrecognized this to be a novel species on the basis of mitochondrialDNA. A denitive species description is provided by Kutscheraet al. (2013).

The H. triserialis colony was established in 2011 with rstgeneration progeny of wild leeches collected in Palo Alto, CA, byUlrich Kutschera, and conveyed to the authors by Dian-Han Kuoand David Weisblat (U.C. Berkeley). Species identity was veriedby sequencing a fragment of cytochrome c oxidase subunit I (COI)amplied with primers 5-GGTCAACAAATCATAAAGATATTGG-3and 5-TAAACTTCAGGGTGACCAAAAAATCA-3.

Inhibition of zygotic transcription

We blocked zygotic transcription with two agents known to beeffective across a wide range of eukaryotic systems. The drug -amanitin (Sigma) was minimally effective in bath application, so0.05 PBS mixed 1:1 with heptane. At stage 7 or older, embryos

The DNA sample in the rst round of selection was 200 ng of

M.W. Schmerer et al. / Developmental Biology 382 (2013) 149159 151were xed with 4% formaldehyde in 0.25 PBS. Fixed embryoswere rinsed in PBS and cleared in glycerol after removal of thevitelline envelope; then examined on a Nikon Eclipse E800microscope with a Diagnostic Instruments Flex 1520 CCD camera.

To trace cell lineage we microinjected cells with 50 mg/mltetramethylrhodamine- or uorescein-dextran-amine (RDA orFDA; Molecular Probes) in 2% fast green. Some xed specimenswere embedded in plastic (POLYBed 812, Polysciences), cut into0.1 mm sections, and imaged on a Zeiss LSM5 Pascal confocalmicroscope. To visualize nuclei, we injected lineages with H2B:RFPmRNA (0.2 mg/ml) encoding human histone H2B fused to reduorescent protein (Kieserman et al., 2008). Proteloblast deriva-tives were scored as ectodermal or mesodermal based on locationin the animal or vegetal hemispheres.

Expression plasmids

A full length GFP-tagged Hau-Pax1 fusion was generated byligating the Pax1 ORF (Schmerer et al., 2009) into a modiedpCRII-TOPO containing eGFP, the multiple cloning region of pCS2+,and an SV40 polyadenylation signal. To produce a dominant-negative construct, the Hau-Pax1 ORF was cloned into pCS107GFPand the Drosophila Engrailed Repressor (EnR) domain ligatedinframe at amino acid 190. EnR was taken from pCS2MT-xSTAT5-EnRmyc (Schmerer et al., 2006) along with a myc tag not utilizedhere. GFP-tagged Pax1:VP16 was generated by swapping the EnR-myc from pGFP-Pax1-EnR with the VP16-myc transactivationdomain of pSP64T-xSTAT5-VP16-myc (Schmerer et al., 2006).The same general procedures were used to generate full length,EnR fusion, and VP16 fusion constructs of genes Hau-Pax3/7A(Woodruff et al., 2007) and -Pax2/5/8 (I.K. Quigley, B.C. Lang, andM. Shankland, unpublished; GenBank accession no. JQ654099).

To abrogate DNA-binding of the GFP:Pax1:EnR protein, weused site-directed mutagenesis (Ke and Madison, 1997) to intro-duce a three amino acid substitution (HSL mutation) convertingpaired domain amino acids 1417 from NGRP to HSRL.

For biochemical analyses, we fused leech paired domain codingsequences to GST by cloning into vector pGEX-4 T-1 (GEHealthcare).

mRNA injections

Expression plasmids were linearized, and mRNAs transcribedand capped in vitro using the SP6 mMessage mMachine kit(Applied Biosystems). Our GFP-tagged mRNAs contained a poly-adenylation signal; untagged Hau-3/7A mRNA did not, and wasadenylated with the Poly(A) Tailing kit (Applied Biosystems) priorto injection.

Pax mRNAs were microinjected at a concentration of 0.20.3 mg/ml in 0.1% Fast Green in the needle. Injection of nls-GFP:EnR mRNA controlled for the non-specic effect of EnR nuclearimport; the latter construct gave much higher protein accumula-tion in vivo, and was injected at 80 g/ml to obtain a comparablelevel of protein. We routinely injected mRNAs into the D blasto-mere of 4-cell embryos, but obtained comparable results injectinguncleaved zygotes or the CD blastomere of 2-cell embryos.

For rescue experiments, the dominant-negative mRNA, rescuemRNA, and 0.2% fast green were injected at 1:4:1 volume ratio.Unless otherwise stated, the concentration of dominant-negativemRNA was 0.2 mg/ml and the rescue mRNA 0.8 mg/ml post-dilution. To assess rescue quantitatively, we compared phenotypesproduced by a given dominant-negative when coinjected with afull length Pax mRNA or the control nls-GFP mRNA. A one-tailedWilcoxon rank sum test was used to determine whetherPax mRNA signicantly increased the number of ectodermal

blastomeres produced, and a one-tailed Fishers test to determineand reprobed with a 1:2000 dilution of monoclonal anti--tubulin(Sigma; clone T0198) using HRP donkey anti-mouse secondary at1:20,000. HRP was detected with the SuperSignal West PicoThe consensus sequence of our site selection analysis was usedas the dsDNA probe. Oligonucleotide 5-CGGTGGGCAATCAAGCGT-GACGACGCCCC-3 was synthesized by Integrated DNA Technolo-gies with a uorophore TYE 665 end label, annealed to itsreverse complement, and gel-puried. Binding reactions wereincubated at room temperature for 20 min, and consisted of 2 lprotein and 20 fmol of dsDNA probe in 20 l volume of TrisPaxbuffer with 1 mM DTT, 2 g poly dIdC, and 4% Ficoll. GST-taggedproteins were expressed in E. coli and puried with MagneGST.GFP-tagged proteins were translated in rabbit reticulocyte lysate(Promega) and introduced into the binding reaction withoutpurication. Binding was analyzed by electrophoresis in 6% poly-acrylamide/0.5 TBE, with uorescent probe DNA imaged on aMolecular Diagnostics Typhoon Scanner (Amersham).

Antibody production and immunoblotting

A polyclonal rabbit serum was generated by Open Biosystemsagainst oligopeptide EPLDLSLKNKFRSQYK, corresponding to thepredicted C-terminus of the Hau-Pax1 cDNA (Schmerer et al.,2009). Serum was afnity-puried by the manufacturer, and usedfor western blots as in Schmerer and Evans (2003). Three embryo-equivalents of protein were loaded onto each lane of a 10% SDS-PAGE gel, separated, and blotted on nitrocellulose. Blots wereincubated in a 1:1000 dilution of afnity-puried anti-Pax1,washed, and incubated in a 1:50,000 dilution of HRP donkeyanti-rabbit secondary. As a loading control, blots were strippeddouble-stranded 5-GCAGTCGACCAAGAGCTCA(N)24AGCACCTGTC-GACGCAG-3. After 1 h at room temperature, GST:Pax1 and itsbound DNA were precipitated with MagneGST, washed ve timesin TrisPax Buffer with 4% Ficoll, and eluted for 1 h at 45 1C in0.1 ml of recovery buffer: 50 mM TrisHCl (pH 8.0), 100 mM Na-acetate, 50 mM EDTA, and 0.5% SDS. Eluted oligos were puried byphenolchloroform extraction/ethanol precipitation, and half thesample PCR amplied with primers corresponding to invariant 5and 3 ends of the oligonucleotide pool. Amplicons were precipi-tated, run on 4% agarose, puried with spin columns, then used asthe DNA sample in the binding reaction of each succeeding round.After nine rounds an aliquot was cloned and sequenced, and the20 sequences aligned manually. The composite binding site wasdepicted using WebLogo 3.0 (http://weblogo.threeplusone.com/).

Electromobility shift assay (EMSA)whether it signicantly increased the fraction of embryos thatdeveloped the normal number of ecto- or mesoteloblasts.

Hau-Pax1 binding site selection

We amplied high afnity Hau-Pax1 binding sites from apopulation of random nucleotide sequences using the PCR-basedselection procedure by Epstein et al. (1994). A GST-tagged Pax1paired domain was expressed in E. coli BL21 DE3, precipitated withparamagnetic glutathione beads (MagneGST kit; Promega), andeluted in recovery buffer. At each round of selection, 10 l of GST:Pax1 fusion protein was incubated with a dsDNA sample in a200 l binding reaction containing TrisPax Buffer (10 mM TrisHCl [pH 7.5], 100 mM KCl, 1 mM EDTA), 4% Ficoll, 0.5 g Poly dIdC,and 1 mM DTT.Chemiluminescent kit (Thermo Fisher).

Results

Inhibition of zygotic transcription

To examine the role of zygotic transcription in early leechdevelopment we injected H. austinensis zygotes with -amanitin atan estimated 1 M intracellular concentration. The vast majority ofembryos (95%; 150/158) developed normally to stage 4b, when cellD cleaved obliquely to produce daughters DM and DNOPQ(cf. Fig. 1A and B). The early cell divisions occurred synchronouslyin amanitin-treated and untreated siblings, and the orientationand symmetry of individual cytokineses appeared to be identicalin the two groups. The few experimental embryos that did notdevelop to stage 4b manifested abnormalities (lysis; failure tocleave; chaotic cleavage) that were also seen at low frequency incontrols.

Amanitin-treated embryos began to undergo abnormal devel-opment during the ensuing cleavages of cells DM and DNOPQ.These two cell lineages normally go through an invariant sequenceof symmetric cleavages, producing meso- or ectoteloblasts respec-tively (Weisblat and Huang, 2001). But 95% (142/150) of amanitin-treated embryos still retained a single large DNOPQ-like cell whenexamined 3050 h postinjection, and 86% retained a single largeDM-like cell (Figs. 2C and 3A). In the other amanitin-treatedembryos the D quadrant blastomeres underwent one or a fewexternally visible cleavages, but none gave rise to the normalcomplement of teloblasts (Fig. 3A).

The effect of -amanitin was dose-dependent. Embryos injectedwith 40 nM -amanitin were on average more likely to undergoexternally visible cleavages (Fig. 3A), although a substantial fraction

still retained a DNOPQ-like cell (68%, 61/90 embryos) and/or a DM-like cell (48%) 3050 h post-injection. Similar results were alsoobtained with a second transcriptional inhibitor, actinomycin D(Fig. 3B). The mesodermal progenitor was more likely than theectoderm to undergo an externally visible cleavage under allconditions (Fig. 3), suggesting that this cell may be less sensitiveto transcriptional perturbation.

Transcription is required for the geometry of proteloblast cleavage

The persistence of DM- and DNOPQ-like cells in amanitin-treated embryos could have resulted from a cessation of celldivision; alternatively, these cells may have experienced asym-metric divisions in which smaller daughters were not readilyvisible. (By analogy, cells DM, DM, and DM cannot be distin-guished on the basis of external appearance in normal embryosdue to the deep location of the dm and dm micromeres.) Todistinguish these possibilities we followed the sequence of prote-loblast cleavages in amanitin-treated embryos, using uorescentlineage tracers as needed.

In the ectoderm we found that the three micromere-producingdivisions leading up to the symmetric cleavage of proteloblastDNOPQ (cf. Fig. 1A and C) occurred with normal timing andgeometry in embryos treated with transcriptional inhibitors(Fig. S1). Cell DNOPQ also cleaved at the appropriate time(710 min of synchronized sibling controls), but the symmetryand orientation of its cytokinesis were aberrant. Unlike thebilaterally symmetric division seen in normal development(Fig. 4A), in embryos treated with 1 M -amanitin the DNOPQ

edermas able fa pe

M.W. Schmerer et al. / Developmental Biology 382 (2013) 149159152Fig. 2. Transcriptional inhibition disrupts proteloblast cleavage. Embryos are viewundergoing bilaterally symmetric teloblast formation. At stage 6a (panel A) the ectodsurrounding the micromere cap (mc). Mesoderm is outside the plane of focus, but hformation. A bilateral pair of N teloblasts and three of the four O/P teloblasts are visiDM- and DNOPQ-like cells. (D) Embryo raised in 200 g/ml actinomycin D showing

into left and right M teloblasts. Scale: 100 m.from animal pole with D quadrant toward the bottom: (A,B) Normal embryosal proteloblast has cleaved repeatedly to produce bilaterally paired N and OPQ cellslso cleaved symmetrically. At stage 7 (panel B) the embryo has completed teloblastrom the animal pole. (C) Embryo injected with 1 M -amanitin showing persistentrsistent DNOPQ-like cell. In this specimen the mesodermal proteloblast has cleaved

M.W. Schmerer et al. / Developmental Biology 382 (2013) 149159 153cleavage was highly asymmetric (Fig. 4B and C) in every case it wasactively watched (n32). The cleavage plane was generally ortho-gonal to the AV axis (Fig. 4B), although sometimes tilted dextrallyoblique such that the smaller daughter was on the left (Fig. 4C).

The small daughter cell produced by these aberrant DNOPQcell divisions was generally yolk-free and consistently located nearthe animal pole at the edge of the micromere cap. Though difcultto follow in unlabeled embryos, this small cell was clearlydistinguished from the dnopqdnopq micromeres by lineagetracer injections (Fig. 4G). In contrast, the larger vegetal daughterinherited both teloplasm and cortex (Fig. 4B and C) and wasreadily visible in unlabeled embryos. Thus, we infer that thepersistent DNOPQ-like cells described in the previous sectionoriginated in most if not all cases as the larger vegetal daughtersof aberrant DNOPQ cleavages.

Amanitin had a similar effect on mesodermal cleavage. Forma-tion of the dm and dm micromeres appeared to be normal, butthe cleavage of proteloblast DM was frequently abnormal.Because the mesoderm cleaves deep within the embryo, weinjected cell DM with RDA in six amanitin-treated embryos andsix sibling controls, then xed these embryos when the controlsreached stage 5 (cf. Fig. 1A). Cell DM manifested its normalsymmetric cleavage in all 6 control embryos (Fig. 4D) and in 1 of6 amanitin-treated embryos. We did not observe an externallyvisible cleavage in the 5 remaining amanitin-treated embryos, butsectioning revealed that their DM cells had in fact undergone

Fig. 3. Histograms showing the number of externally visible ectodermal andmesodermal in embryos treated with pharmacological inhibitors of transcription.A color key is provided at the bottom: teal and green bars represent embryos thatproduced the normal complement of ecto- and/or mesoteloblasts; red, orange, andyellow bars represent embryos with a reduced number; dark blue bars representembryos with an excess. The number of ectoteloblasts normally visible at this stageranges from 6 to 8 due to individual variation in the internalization of Q teloblasts.Specimens were scored 3050 h after the onset of treatment, and total numbers areshown at the right. Zygotes that did not develop normally to stage 4b (o5%) arenot included.highly asymmetric divisions with a cleavage plane orientedorthogonal to the AV axis (Fig. 4E). These aberrant DM cleavagesproduced a small yolk-free animal daughter beneath the micro-mere cap and a large vegetal daughter that inherited bothteloplasm and cortex (Fig. 4E). Thus, we infer that the persistentDM-like cells of the previous section originated in most if not allcases as the large vegetal daughters of aberrant DM cleavages. Incertain other experiments we observed amanitin-treated DMcells cleave with a less extreme asymmetry (Fig. 4F), and whenthis occurred the smaller daughter cell was consistently onthe left.

Injection of -amanitin has been reported to produce a some-what different phenotype in embryos of the congeneric leech H.triserialis (Bissen and Weisblat, 1991). We repeated those experi-ments here using a closely related strain of H. triserialis (Fig. S2),and found that transcriptional inhibition frequently disruptedthe symmetry of proteloblast cleavage in this second species aswell (Fig. S3).

Later development of transcriptionally inhibited embryos

The persistent DM- and DNOPQ-like cells continued to dividefor 34 days after amanitin treatment. Each cell generated a linearcolumn of small, uniformly sized daughter cells (Fig. 4HJ) thatoutwardly resembled the blast cell bandlets seen during normalleech development. These columns emerged from the teloplasm ofthe parent blastomere, and were attached distally to the micro-mere cap (Fig. 4I). However, the mesodermal and ectodermalcolumns did not merge to form a bilayered germinal band(Fig. 4H), and no subsidiary mitoses were observed within thetimeframe of our observation. All amanitin-treated embryos diedwithout undergoing gastrulation.

Dominant-negative knockdown of Pax gene function

While investigating the function of Pax genes in leech devel-opment we discovered that dominant-negative Pax constructs canproduce cleavage defects similar to those observed with -amanitin. We focus here on three H. austinensis Pax genes, all ofwhich are present as maternal RNAs and prevalent in D quadrantblastomeres leading up to proteloblast cleavage: Hau-Pax3/7A(Woodruff et al., 2007); Hau-Pax1 (Schmerer et al., 2009); andHau-Pax2/5/8 (I.K. Quigley, B.C. Lang, M. Shankland, unpublishedresults).

For each of these genes a dominant repressor construct wasgenerated by fusing the Drosophila EnR domain to the 3 end of theleech paired domain (Conlon et al., 1996). GFP was fused to the 5end of both dominant-negative and full length constructs. Follow-ing mRNA injection, GFP uorescence was used to verify thetranslation and nuclear localization of the fusion protein in thedeveloping embryo.

Hau-Pax2/5/8: Injecting embryos with 0.2 mg/ml GFP:Hau-Pax2/5/8:EnR mRNA had no obvious effect on the timing orgeometry of the earliest embryonic cell divisions, but blocked ortruncated the symmetric cleavages of the proteloblasts in everycase (Fig. 5G). Ninety-two percent (72/78) of these embryosretained a large DNOPQ-like blastomere at 2830 h post-injection,and 32% (25/78) retained a large DM-like blastomere (Fig. 5A). In anumber of cases we observed cell DNOPQ undergo highlyasymmetric cleavages similar to those seen with amanitin treat-ment, and the persistent DM- and DNOPQ-like cells often went onto generate linear columns of small, yolk-free daughters. Given thesimilarities in phenotype, these ndings suggested that GFP:Hau-Pax2/5/8:EnR was repressing one or more of the zygotic genes that

are normally required for symmetric proteloblast cleavage.

M.W. Schmerer et al. / Developmental Biology 382 (2013) 149159154Embryos injected with full length GFP:Hau-Pax2/5/8 cleavednormally to produce the usual complement of teloblasts (Fig. 5Dand G), as did embryos injected with the Hau-Pax2/5/8 paireddomain fused to transcriptional activator VP16 (Fig. 5G; seeSadowski et al., 1988). Normal cleavage was also observed(Fig. 5G) in embryos expressing GFP fused to the EnR (3) and anuclear localization signal (5), demonstrating that cleavage wasunaffected by the mere entry of EnR into the nucleus. Hence, thecleavage defects described above apparently resulted from physi-cal association of EnR with the paired domain.

Hau-Pax1: Injection of 0.2 mg/ml GFP:Hau-Pax1:EnR mRNAalso blocked or truncated the normal sequence of ectodermalcleavages (Fig. 5B and G), although the phenotype was typicallyless severe. Twenty-three percent (18/80) of these embryosretained a single large DNOPQ-like blastomere, and 89% showeda clear reduction in the number of ectoteloblasts produced. Thesymmetric cleavage of the mesodermal proteloblast was some-times blocked as well, but this effect was relatively rare inembryos injected with GFP:Hau-Pax1:EnR (Fig. 5G; see alsoFig. 6). Both germ layers developed normally in embryos injectedwith full length GFP:Hau-Pax1 (Fig. 5E), as well as the vastmajority of embryos injected with GFP:Hau-Pax1:VP16.

Hau-Pax3/7A: Injection of 0.2 mg/ml GFP-Hau-Pax3/7A:EnRmRNA produced cleavage defects similar to those seen with our

Fig. 4. Transcriptional inhibition altered the geometry of proteloblast cleavage: (AC) Elocalizes to the teloplasm. Panel A is an animal view of the normal symmetric cleavagamanitin, and show asymmetric animal (an) and vegetal (vg) daughters. The micromerelabeled with RDA. Panel D is a vegetal view of the normal symmetric cleavage. Panel Easymmetrically to produce a small animal daughter (an) beneath the micromere cap. Pancleaved with moderate asymmetry; the left side daughter (L) is smaller than the right (great-granddaughter DNOPQ reinjected with FDA. DNOPQ underwent an aberrant asclearly distinct from the singly labeled (red) dnopqdnopq micromeres (marked 13).the doubly labeled vegetal daughter is visible at the right. (H) Mesodermal (red) and eproducing columns of yolk-free daughter cells (arrows). (I,J) High magnication of an ecand nuclear morphology by H2B:RFP protein (red). A faintly labeled cluster of DNOPQ-d100 m; G, I, J 20 m.other Pax:EnR constructs (Fig. 5C). However, comparable abnorm-alities were also observed following the injection of full lengthHau-Pax3/7A mRNA (Fig. 5F), with or without a GFP tag. Theseobservations could be explained if full length Hau-Pax3/7Amimicked the EnR constructs by acting as a transcriptionalrepressor, an interpretation supported by our nding that theHau-Pax3/7A paired domain did not elicit cleavage defects whenfused to the VP16 activator (Fig. 5G).

In sum, these dominant-negative studies suggested that theHau-Pax2/5/8 and/or -Pax1 genes could play a role in regulatingthe symmetry of proteloblast cleavage during normal leech devel-opment. Proteloblast cleavage was also altered by our Hau-Pax3/7Adominant-negative, but control experiments suggested that thiswas an artifact, and possibly the result of off-target effects.

Functional rescue of dominant-negative phenotypes

To elucidate the genetic specicity of these dominant-negativephenotypes we tried to rescue proteloblast cleavage by coexpres-sion of full length Pax proteins (see section Methods). For eachEnR construct, we compared the phenotypic outcome of usingGFP:Hau-Pax2/5/8, GFP:Hau-Pax1, or the control nls-GFP as therescue mRNA. The range of phenotypes obtained using nls-GFP torescue served as the baseline for statistical analyses. Results are

mbryos in which proteloblast DNOPQ was labeled with RDA, which preferentiallye. Panels B (dorsal view) and C (animal view) are embryos treated with 1 M -cap (mc) is marked for orientation. (DF) Embryos in which proteloblast DM wasis an optical section of an embryo treated with 1 M -amanitin: DM has cleavedel F is a vegetal view of a different embryo treated with 1 M -amanitin: DM hasR). (G) Amanitin-treated embryo in which cell DNOPQ was injected with RDA, andymmetric cleavage, and its doubly labeled (yellow) small animal daughter (an) isImage is a collapsed Z series of horizontal optical sections, and a glancing section ofctodermal (green) blastomeres continue to divide after 3 d of amanitin treatment,todermal cell column: cell morphology is delineated by FDA lineage tracer (green);erived micromeres (asterisk) is visible at the columns leading end. Scales: AF, H

M.W. Schmerer et al. / Developmental Biology 382 (2013) 149159 155summarized in Table 1, and a subset of data presented graphicallyin Fig. 6.

GFP:Hau-Pax2/5/8: Coinjection of GFP:Hau-Pax2/5/8 mRNAappeared to rescue the symmetric cleavage of the mesodermalproteloblast under several different conditions (Figs. 6 and S4). Forexample, cell DM only cleaved normally in one-quarter of theembryos injected with a combination of 0.2 mg/ml GFP:Hau-Pax2/5/8:EnR and 0.8 mg/ml nls-GFP, but substituting GFP:Hau-Pax2/5/8as the rescue construct gave a highly signicant increase(po0.0005) in the fraction of embryos that produced an overtly

Fig. 5. In vivo expression of dominant-negative Pax constructs. Cells labeled as in Fig.Embryos injected with mRNAs encoding full length Pax proteins. Scale: 100 m. (G) Hblastomeres in embryos injected with Pax gene constructs. Specimens scored 2830 h atotal numbers shown at the right. The white bar in the Pax3/7 A:EnR histogram represe

Fig. 6. Histograms showing the number of externally visible ectodermal and mesodermmRNAs. Dominant-negative mRNAs (top) and full length rescue mRNAs (left) were injectscored 2830 h after mRNA injection. A color key is provided at the bottom (see legendnormal pair of M teloblasts (Table 1). Coinjection of GFP:Hau-Pax2/5/8 also gave a substantial number of embryos (blue bar in Fig. 6)in which the mesodermal proteloblast underwent multiple roundsof symmetric cleavage, producing an array of 36 abnormallysmall mesoteloblasts (Fig. S6B). Hence, the probability that cell DM would manifest at least one symmetric cleavage was roughlytripled by coinjection of full length Hau-Pax2/5/8 (Fig. 6).

GFP:Hau-Pax2/5/8 also increased the likelihood of normalmesodermal cleavage in embryos when coinjected with the GFP:Pax3/7A:EnR or GFP:Hau-Pax1:EnR dominant-negatives (Table 1).

2: (AC) Embryos injected with mRNAs encoding Pax:EnR fusion proteins. (DF)istograms showing the number of externally visible ectodermal and mesodermalfter mRNA injection. A color key is provided at the bottom (see legend, Fig. 3), andnts a single embryo in which cell D did not cleave.

al blastomeres in embryos coinjected with dominant-negative and full length Paxed at a 1:4 ratio; nls-GFP was employed as a control rescue mRNA. Specimens were, Fig. 3), and total numbers shown at the right.

ome

0 77nnnn 781 53nnn 66

ativrol

M.W. Schmerer et al. / Developmental Biology 382 (2013) 149159156However, only the former effect was statistically signicant(po0.01).

In contrast to its clearcut effect on mesodermal development,coinjection of full length Hau-Pax2/5/8 had no apparent effect onectodermal cleavage (Fig. S4A). There was no signicant increasein the number of ectodermal blastomeres produced when thisrescue construct was combined with any of our three dominant-negatives (Table 1).

GFP:Hau-Pax1: Coinjection of GFP:Hau-Pax1 was able torescue both mesodermal and ectodermal cleavage defects, anddid so in combination with any of our three dominant-negativeconstructs (Fig. 6). When coinjected with GFP:Hau-Pax1:EnR, fulllength Hau-Pax1 gave a highly signicant increase (po0.0001) inboth the mean number of ectodermal blastomeres produced andthe percentage of embryos displaying the normal complement ofectoteloblasts (Table 1). The GFP:Hau-Pax1:EnR dominant-negative had a relatively weak effect on mesodermal cleavage,but this was also alleviated to a signicant degree (po0.05) bycoinjection of GFP:Hau-Pax1 (Table 1). The cleavage pattern of

Table 1Rescue of dominant-negative phenotypes by full-length Pax proteins.

Dominant-negative mRNA Rescue mRN Ectodermal blast

GFP:Hau-Pax1:EnR (0.2 mg/ml) GFP:Hau-Pax1 5.571.7nnnn

GFP:Hau-Pax2/5/8 2.471.3nls-GFP 2.471.8

GFP:Hau-Pax2/5/8:EnR (0.2 mg/ml) GFP:Hau-Pax1 1.270.5GFP:Hau-Pax2/5/8 1.470.8nls-GFP 1.270.7

GFP:Hau-Pax2/5/8:EnR (2 g/ml) GFP:Hau-Pax1 8.070.2nnnn

GFP:Hau-Pax2/5/8 3.071.4nls-GFP 3.171.6

GFP:Hau-Pax3/7A:EnR (0.2 mg/ml) GFP:Hau-Pax1 2.972.2n

GFP:Hau-Pax2/5/8 1.470.7nls-GFP 2.171.5

Rescue mRNAs were injected at four-times the concentration of the dominant-negsignicantly increased by coinjecting the full-length Pax protein instead of the cont

n po0.05.nn po0.01.nnn po0.0005.nnnn po0.0001.these rescued embryos was indistinguishable from normal inmany cases (Fig. S4C).

When GFP:Hau-Pax1was coinjected at a 4:1 ratio with 0.2 mg/ml GFP:Hau-Pax2/5/8:EnR, there was a highly signicant rescue(po0.0001) of mesodermal cleavage but little or no alleviation ofectodermal defects (Table 1). Given that ectodermal cleavage wasmore sensitive to disruption following a wide range of experi-mental perturbations, we examined the interaction of GFP:Hau-Pax1 and GFP:Hau-Pax2/5/8:EnR when both mRNAs had beendiluted one hundred-fold. At this lower concentration, full lengthHau-Pax1 gave a highly signicant rescue (po0.0001) of theectodermal cleavage pattern as well (Table 1; Figs. 6, S4D).

In addition, coinjection of GFP:Hau-Pax1 gave a modest butstatistically signicant improvement in both ectodermal andmesodermal cleavage when coinjected with the dominant-negative GFP:Pax3/7A:EnR (Table 1).

DNA binding of Pax gene constructs

We compared the DNA binding properties of the Pax constructsemployed in these functional studies. First we used a GST:Pax1paired domain to select sequences from a random pool of double-stranded oligonucleotides, cloning selected sequences after ninerounds of serial enrichment (see section Methods). The selectedoligonucleotides shared the sequence motif CA(-)6CGTGAC (Fig. S5A),and the consensus was used as a DNA probe for EMSAs. The Hau-Pax1 paired domain gave a strong gel shift when tagged with GFP orGST (Fig. S5B and C), as did the GFP:Pax1:EnR and GFP:Pax1:VP16constructs (Fig. S5B). In short, DNA binding was not compromised inany of the constructs used to analyze function in vivo. The Hau-Pax2/5/8 and -Pax3/7A paired domains also bound this probe (Fig. S5C).

To determine whether DNA binding was critical for thedominant-negative phenotype, we introduced three amino acidsubstitutions (HSL mutation, see section Methods) into thepaired domain of GFP:Hau-Pax1:EnR. These substitutions targeta highly conserved protein:DNA interface (Xu et al., 1995), andgreatly reduced DNA binding in EMSAs (Fig. S5B). Ninety-sixpercent (24/25) of embryos injected with GFP:Pax1HSL:EnRmRNAproduced a normal complement of teloblasts, suggesting that ourdominant-negatives had to bind DNA in order to disrupt thesymmetry of proteloblast cleavage.

Developmental time-course of Hau-Pax1 protein expression

1 25 119

100nnnn 100 240 92 255 96 82

17n 96nn 470 97nn 393 72 31

e in all combinations. Statistical tests calculated whether a given parameter wasnls-GFP. Signicant results are marked as follows:res (mean7S.D.) % Normal ectoderm % Normal mesoderm N

73nnnn 97n 714 95 797 86 74To examine Hau-Pax1 protein expression we generated apolyclonal antiserum against an oligopeptide corresponding tothe C terminus. Afnity-puried anti-Hau-Pax1 recognized bandsof 130 and 110 kDa in protein extracts of embryos (Fig. S6). Neitherband was detected by preimmune serum, and staining wasblocked by oligopeptide preincubation. Both bands showed arelatively constant abundance throughout embryogenesis (Fig.S6), being prevalent in oocytes, uncleaved zygotes, and earlycleavage-stage embryos (stages 26). Hence, the embryo receivesa signicant contribution of maternal Hau-Pax1 protein, and thelevel of Hau-Pax1 remains stable leading up to the symmetricproteloblast cleavages. This anti-Hau-Pax1 serum did not proveeffective for immunohistochemistry.

Discussion

Maternal and zygotic factors in the regulation of embryonic celllineage

Maternal gene products orchestrate the earliest embryonic celldivisions in a wide variety of animal taxa (Newport and Kirschner,1982; Edgar and Schubiger, 1986; Edgar et al., 1994), but eventually

M.W. Schmerer et al. / Developmental Biology 382 (2013) 149159 157give way to zygotic regulation as development proceeds (OFarrellet al., 1989). Consistent with this scheme we nd that leechembryos treated with transcriptional inhibitors undergo normalearly development, then manifest an abrupt requirement for denovo transcription during the cleavage of the DM and DNOPQproteloblasts. Inhibitors had little or no effect on the timing ofthese two cell divisions, arguing that cell cycle progression wasunder the control of maternal factors, but the placement of theproteloblasts cleavage furrows was strongly inuenced by zygoticgene expression.

The strongest and most consistent cleavage defects wereobserved in H. austinensis embryos treated with 1 M -amanitin,a concentration known to eliminate nearly all RNA polymerase II-mediated transcription in vitro (Lindell et al., 1970). In theseembryos the DM and DNOPQ cells rarely manifested theirnormal cleavages, which are bilaterally symmetric, but ratherdisplayed highly asymmetric divisions with an orientation roughlyorthogonal to the normal cleavage. This aberrant cleavage geome-try was strikingly similar in the mesodermal and ectodermallineages, suggesting that the two proteloblasts use comparablemechanisms and possibly some of the same gene products (seebelow) to regulate cleavage geometry during normaldevelopment.

The proteloblasts DM and DNOPQ are the rst cells in the leechembryo to display bilaterally symmetric cleavage, and their transitionto bilateral symmetry is a key event in morphogenesis. Our presentndings suggest that the proteloblasts require zygotic gene expres-sion to undertake this developmental transition, raising the possibi-lity that they simply recapitulate the micromere-producing celldivisions of their parents in the absence of transcription. Forexample, the parent cell DNOPQ normally divides to produce thelarge yolk-rich proteloblast and a small yolk-free animal micromere,a cytokinesis that resembles the aberrant cleavage of DNOPQ inembryos treated with transcriptional inhibitors. The same is also trueof the mesodermal lineage. Indeed, the fact that proteloblastsgenerate linear columns of small, uniformly sized daughter cells inamanitin-treated embryos could mean that they are doomed to serialrepetition of the parental cell division.

It should, however, be noted that columns of small uniformlysized daughter cells are also observed during normal leech devel-opment, i.e. the blast cell bandlets generated by the meso- andectoteloblasts (Weisblat and Huang, 2001). Thus, an alternativeinterpretation of our ndings is that proteloblasts lacking zygotictranscription may mature into teloblasts precociously, and indoing so skip over the symmetric cleavages that are part of normaldevelopment. In theory one could distinguish these two models bycomparing markers of micromere vs. blast cell differentiation.However, it would be difcult to make such a distinction withany condence relying solely on the morphological data at hand.

Independent of these models, our ndings suggest that theleech embryos normal transition from spiral to symmetric clea-vage is brought about either directly or indirectly by the de novosynthesis of one or more zygotic gene products. The regulation ofembryonic cell divisions has been studied extensively in C. elegansand Drosophila embryos (reviewed by Siller and Doe, 2009), and acentral theme to come out of those studies is that the placementand orientation of the mitotic spindle are controlled by evolutio-narily conserved pathways including the PAR cell polarity proteins(Goldstein and Macara, 2007) and receptor-independent G proteinsignaling (Hampoelz and Knoblich, 2004). These pathways worktogether to regulate the cytoplasmic motors that engage and pullon astral microtubules (Moore and Cooper, 2010). It could be thatleech proteloblasts normally shift from spiral to symmetric clea-vages because they or their lineal precursors begin to expressgene products that reset the balance between different regulatory

pathways.Pax genes and proteloblast cleavage

Our dominant-negative studies suggest that there is at leastone transcriptional event required for the symmetry of proteoblastcleavage that is critically dependent upon Pax gene transcriptionfactors. The cleavage defects produced by dominant-negative Paxconstructs were on average weaker than those seen with -amanitin, but encompassed a range of developmental phenotypesthat was largely indistinguishable. Moreover, the dominant-negative phenotype was alleviated by a mutation that blocksDNA binding, and could be rescued by coexpression of full lengthPax proteins.

We were surprised that the genetic specicity of our rescueexperiments differed so markedly in ectodermal and mesodermallineages. In the ectoderm, coexpression of Hau-Pax1 reduced oreliminated the cleavage defects produced by all three dominant-negative constructs. We interpret this rescue as evidence that Hau-Pax1 promotes the transcription of zygotic gene(s) required forthe symmetric cleavage of DNOPQ, and infer that endogenousHau-Pax1 very likely activates those same gene(s) during normaldevelopment. The prevalence of Hau-Pax1 RNA (Schmerer et al.,2009) and protein in the stages leading up to proteloblast cleavageis consistent with this interpretation.

In contrast, Hau-Pax2/5/8 gave no detectable rescue of ectoder-mal cleavage within the context of our assay. Although a negativeresult, this observation suggests that endogenous Hau-Pax2/5/8does not contribute to ectodermal cleavage in vivo. It should benoted that Hau-Pax2/5/8:EnR gave the strongest and most consis-tent ectodermal cleavage defects of any of our dominant-negativeconstructs, suggesting that this protein can compete with Hau-Pax1 for DNA binding and elicit off-target repression. But fulllength Hau-Pax2/5/8 seems unable to replace Hau-Pax1 as atransactivator of those genes.

The genetic specicity of mesodermal rescue was decidedlydifferent, with coinjection of either Hau-Pax2/5/8 or -Pax1 beingsufcient to rescue the dominant-negative phenotype. This func-tional redundancy could be an artifact of our rescue paradigm, or itmay indicate that both Hau-Pax2/5/8 and -Pax1 contribute to theregulation of mesodermal cleavage during normal development. Ifcorrect, this would mean that the symmetric cleavages of theectodermal and mesodermal proteloblasts depend to at least somedegree upon distinct transcriptional events. In further support ofthat idea, we have found a 12 h interval in stage 4b whenectodermal cleavage remains sensitive to -amanitin injectionwhile the mesoderm no longer responds (M. Shankland, unpub-lished results). Genetic redundancy in the control of mesodermalcleavage could also help to explain why the DM cleavage wasconsistently less sensitive to perturbation by either pharmacolo-gical or molecular reagants.

It is interesting that our dominant-negative Pax constructs hadno discernable effect during the 12 h of development leading up toproteloblast cleavage, even though the embryo inherits Hau-Pax1protein from the oocyte. It is possible that this store of maternalprotein is sufcient to regulate proteloblast cleavage, and is simplynot employed as a transcriptional regulator during the earliest cellcycles. In this regard, it is noteworthy that antisense morpholinoknockdown of Hau-Pax1 has no effect on early cleavages, althoughlater abnormalities have been observed (Schmerer and Shankland,unpublished results). Still, we cannot rule out the possibility thatleech Pax proteins have other early developmental functions thatwere overlooked here due to our focus on cell lineage.

Comparison to other spiralians

The bilaterally symmetric division of mesodermal progenitor cell

4d (homolog of leechs DM) and the role of that cleavage in

specify the D quadrant (Lambert and Nagy, 2001, 2003; Henry andPerry, 2008; but also see Amiel et al., 2013), it is largely unknown how

Henry, J.J., Perry, K.J., 2008. MAPK activation and the specication of the D quadrantin the gastropod mollusc Crepidula fornicata. Dev. Biol. 313, 181195.

single-tube megaprimer PCR method. Nucleic Acids Res. 25, 33713372.Kutschera, U., Langguth, H., Kuo, D.-H., Weisblat, D.A., Shankland, M., 2013.

M.W. Schmerer et al. / Developmental Biology 382 (2013) 149159158that specication leads to symmetric cleavage of cell 4d.In the leech H. austinensis, our ndings indicate that zygotic

transcription mediated by Hau-Pax1 and/or -Pax2/5/8 plays acritical role in this process. Bissen and Weisblat (1991) examinedthe effect of -amanitin another leech species, H. triserialis, andwhile they also observed abnormalities in teloblast formation thoseauthors reported that only 13% of amanitin-treated embryos under-went abnormal development during the two hours leading up tostage 5, i.e. when DM and DNOPQ normally cleave. We repeatedthose experiments here on a closely related strain of H. triserialis,and in our hands transcriptional inhibition had a variable butnonetheless dramatic effect on the initial symmetry of proteloblastcleavage. We therefore conclude that zygotic gene expression doesregulate the transition to bilateral symmetry in both leech species,although the role of Pax genes in H. triserialis is still unknown.

There are only a few pertinent studies frommore distantly relatedspiralians. It has been reported that actinomycin D has no effect onthe cleavage of polychaete (Guerrier, 1971) or molluscan (Newrockand Raff, 1975) embryos, but those studies cannot be deemedconclusive since the authors failed to insure that transcription wasinhibited before the symmetric division of cell 4d. And there are noreports on the developmental function of Pax or Pax2/5/8 genes inother spiralians, although Has-Pax258 has been shown to be amaternal gene product in abalone (OBrien and Degnan, 2003).

Despite this dearth of comparative data, the evolutionaryhistory of the Pax gene family is interesting in light of thefunctional studies reported here. Pax genes have only been foundin taxa that develop by spiral cleavage or evolved from spiralianancestors (Schmerer et al., 2009), suggesting that the Pax familyoriginated near the base of the spiralian radiation. Moreover,intron/exon organization argues that Pax arose by duplicationand divergence of a Pax2/5/8 gene (Schmerer et al., 2009), in whichcase the functional redundancy of Hau-Pax1 and -Pax2/5/8 inleech mesoderm could reect an ancestral role that predated geneduplication. One can only speculate as to how the earliestspiralians managed to retain bilateral morphogenesis while evol-ving a cleavage pattern dominated by rotational divisions. But theability of Hau-Pax1 and/or -Pax2/5/8 to bring about a spiral-to-bilateral cleavage transition in the leechs 4d lineage may offerinsight into this question.

In contrast to the mesoderm, only a subset of spiralians,primarily annelids, initiate the bilateral symmetry of their ecto-dermal development through the symmetric cleavage of a largeembryonic progenitor cell (Meyer and Seaver, 2011). Our func-tional data indicate that H. austinensis employs Pax-mediatedzygotic transcription to govern the cleavage symmetry of bothmesodermal and ectodermal proteloblasts, and may in fact use oneof the same transcription factors, Pax1, in both cells. Thismechanistic similarity could bespeak a common evolutionaryorigin: for example, the transcriptional mechanism for regulatingcleavage symmetry may have originated in the spiralian meso-derm, and later been coopted to the ectoderm in an ancestor of theleech. However, these developmental phenomena will need to beinvestigated in a broader phylogenetic context before one canmeaningfully address evolutionary hypotheses of this sort.

Acknowledgments

The authors thank Muriel Umbhauer and John Wallingfordsegregating the left and right halves of the mesoderm are nearlyuniversal hallmarks of spiralian development (Henry and Martindale,1999). But while there is conservation in the molecular pathways thatfor contributing plasmids, Erin Murphy for the site-directedDescription of a new leech species from North America, Helobdella austinensisn. sp. (Hirudinea: Glossiphoniidae), with observations on its feeding behaviour.Zoosyst. Evol. 89, 239246.

Lambert, J.D., Nagy, L.M., 2001. MAPK signaling by the D quadrant embryonicorganizer of the mollusc Ilyanassa obsoleta. Development 128, 4556.

Lambert, J.D., Nagy, L.M., 2003. The MAPK cascade in equally cleaving spiralianembryos. Dev. Biol. 263, 231241.

Lang, D., Powell, S.K., Plummer, R.S., Young, K.P., Ruggeri, B.A., 2007. PAX genes:roles in development, pathophysiology, and cancer. Biochem. Pharmacol. 73,114.

Lindell, T.J., Weinberg, F., Morris, P.W., Roeder, R.G., Rutter, W.J., 1970. SpecicHolton, B., Wedeen, C.J., Astrow, S.H., Weisblat, D.A., 1994. Localization of poly-adenylated RNAs during teloplasm formation and cleavage in leech embryos.Rouxs Arch. Dev. Biol. 204, 4653.

Kieserman, E.K., Glotzer, M., Wallingford, J.B., 2008. Developmental regulation ofcentral spindle assembly and cytokinesis during vertebrate embryogenesis.Curr. Biol. 18, 116123.

Ke, S.H., Madison, E.L., 1997. Rapid and efcient site-directed mutagenesis bymutagenesis protocol, and Charlie Hall and Todd Evans for advice.We would also like to thank Ulrich Kutschera, David Weisblat, andDian-Han Kuo for the H. triserialis specimens used in this study.Lastly, we are indebted to Jonathan Henry for commenting on themanuscript, and to Jeff Gross for feedback on the manuscript andthe use of his confocal microscope. This study was funded by NSFgrants IBN-0415732 and IOS-0718441 to M.S. and a U.T. Austinundergraduate research fellowship to R.W.N.

Appendix A. Supporting information

Supplementary data associated with this article can be found inthe online version at: http://dx.doi.org/10.1016/j.ydbio.2013.07.015.

References

Amiel, A.R., Henry, J.Q., Seaver, E.C., 2013. An organizing activity is required for headpatterning and cell fate specication in the polychaete annelid Capitella teleta:new insights into cellcell signaling in Lophotrochozoa. Dev. Biol. 379, 107122.

Balczarek, K.A., Lai, Z.C., Kumar, S., 1997. Evolution of functional diversication ofthe paired box Pax DNA-binding domains. Mol. Biol. Evol. 14, 829842.

Bely, A.E., Weisblat, D.A., 2006. Lessons from leeches: a call for DNA barcoding inthe lab. Evol. Dev. 8, 491501.

Bergmann, D.C., Lee, M., Robertson, B., Tsou, M.F., Rose, L.S., Wood, W.B., 2003.Embryonic handedness choice in C. elegans involves the G protein GPA-16.Development 130, 57315740.

Bissen, S.T., Weisblat, D.A., 1991. Transcription in leech: mRNA synthesis is requiredfor early cleavages in Helobdella embryos. Dev. Biol. 146, 1223.

Conlon, F.L., Sedgwick, S.G., Weston, K.M., Smith, J.C., 1996. Inhibition of Xbratranscription activation causes defects in mesodermal patterning and revealsautoregulation of Xbra in dorsal mesoderm. Development 122, 24272435.

Dunn, C.W., Hejnol, A., Matus, D.Q., Pang, K., Browne, W.E., Smith, S.A., Seaver, E.,Rouse, G.W., Obst, M., Edgecombe, G.D., Sorensen, M.V., Haddock, S.H., Schmidt-Rhaesa, A., Okusu, A., Kristensen, R.M., Wheeler, W.C., Martindale, M.Q., Giribet,G., 2008. Broad phylogenomic sampling improves resolution of the animal treeof life. Nature 452, 745749.

Edgar, B.A., Schubiger, G., 1986. Parameters controlling transcriptional activationduring early Drosophila development. Cell 44, 871877.

Edgar, L.G., Wolf, N., Wood, W.B., 1994. Early transcription in Caenorhabditis elegansembryos. Development 120, 443451.

Epstein, J., Cai, J., Glaser, T., Jepeal, L., Maas, R., 1994. Identication of a Pax PDrecognition sequence and evidence for DNA-dependent conformationalchanges. J. Biol. Chem. 269, 83558361.

Gline, S.E., Nakamoto, A., Cho, S.-J., Chi, C., Weisblat, D.A., 2011. Lineage analysis ofmicromere 4d, a super-phylotypic cell for Lophotrochozoa, in the leechHelobdella and the sludgeworm Tubifex. Dev. Biol. 353, 120133.

Goldstein, B., Macara, I.G., 2007. The PAR proteins: fundamental players in animalcell polarization. Dev. Cell 13, 609622.

Guerrier, P., 1971. A possible mechanism of control of morphogenesis in the embryoof Sabellaria alveolata (Annelide polychaete). Exp. Cell Res. 67, 215218.

Hampoelz, B., Knoblich, J.A., 2004. Heterotrimeric G proteins: new tricks for an olddog. Cell 119, 453456.

Henry, J.J., Martindale, M.Q., 1999. Conservation and innovation in spiraliandevelopment. Hydrobiologia 402, 255265.inhibition of nuclear RNA polymerase II by -amanitin. Science 170, 447449.

Matus, D.Q., Pang, K., Daly, M., Martindale, M.Q., 2007. Expression of Pax genefamily members in the anthozoan cnidarians, Nematostella vectensis. Evol. Dev.9, 2538.

Meyer, N., Seaver, E.C., 2011. Cell lineage and fate map of the primary somatoblastof the polychaete annelid Capitella teleta. Int. Comp. Biol. 50, 756767.

Moore, J.K., Cooper, J.A., 2010. Coordinating mitosis with cell polarity: molecularmotors at the cortex. Semin. Cell Dev. Biol. 21, 283289.

Newport, J., Kirschner, M., 1982. A major developmental transition in early Xenopusembryos: I. Characterization and timing of cellular changes at the midblastulastage. Cell 30, 675686.

Newrock, K.M., Raff, R.A., 1975. Polar labe specic regulation of translation inembryos of Ilyanassa obsoleta. Dev. Biol. 42, 242261.

OBrien, E.K., Degnan, B.M., 2003. Expression of Pax258 in the gastropod statocyst:insights into the antiquity of metazoan geosensory organs. Evol. Dev. 5,572578.

OFarrell, P.H., Edgar, B.A., Lakich, D., Lehner, C.F., 1989. Directing cell divisionduring development. Science 246, 635640.

Sadowski, I., Ma, J., Triezenberg, S., Ptashne, M.., 1988. GAL4-VP16 is an unusuallypotent transcriptional activator. Nature 335, 563564.

Schmerer, M., Evans, T., 2003. Primitive erythropoesis is regulated by SMAD-dependent signaling in postgastrulation mesoderm. Blood 102, 31963205.

Schmerer, M.W., savage, R.M., Shankland, M., 2009. pax: a novel family oflophotrochozoan pax genes. Evol. Dev. 11, 689696.

Schmerer, M., Torregroza, I., Pascal, A., Umbhauer, M., Evans, T., 2006. STAT5 acts asa repressor to regulate early embryonic erythropoesis. Blood 108, 29892997.

Seaver, E.C., Shankland, M., 2000. Leech segmental repeats develop normally in theabsence of signals from either anterior or posterior segments. Dev. Biol. 224,339353.

Siddall, M.E., Borda, E., 2003. Phylogeny and revision of the leech genus Helobdella(Glossiphoniidae) based on mitochondrial gene sequences and morphologicaldata and a special consideration of the triserialis complex. Zool. Scr. 32, 2333.

Siller, K.H., Doe, C.Q., 2009. Spindle orientation during asymmetric cell division.Nat. Cell Biol. 11, 365374.

Sulston, J.E., Schierenberg, E., white, J.G., Thomson, J.N., 1983. The embryonic celllineage of the nematode Caenorhabditis elegans. Dev. Biol. 100, 64119.

Wang, Q., Fang, W.-H., Krupinski, J., Kumar, S., Slevin, M., Kumar, P., 2008. Pax genesin embryogenesis and oncogenesis. J. Cell Mol. Med. 12, 22812294.

Weisblat, D.A., Huang, F.Z., 2001. An overview of glossiphoniid leech development.Can. J. Zool. 79, 218232.

Woodruff, J.B., Mitchell, B.J., Shankland, M., 2007. Hau-Pax3/7 A is an early markerof leech mesoderm involved in segmental morphogenesis, nephridial develop-ment, and body cavity formation. Dev. Biol. 306, 824837.

Xu, W., Rould, M.A., jun, S., Desplan, C., Pabo, C.O., 1995. Crystal structure of a paireddomainDNA complex at 2.5 resolution reveals structural basis for Paxdevelopmental mutations. Cell 80, 639650.

M.W. Schmerer et al. / Developmental Biology 382 (2013) 149159 159

Developmental transition to bilaterally symmetric cell divisions is regulated by Pax-mediated transcription in embryos of...IntroductionMethodsEmbryosInhibition of zygotic transcriptionLineage tracing and histologyExpression plasmidsmRNA injectionsHau-Pax1 binding site selectionElectromobility shift assay (EMSA)Antibody production and immunoblotting

ResultsInhibition of zygotic transcriptionTranscription is required for the geometry of proteloblast cleavageLater development of transcriptionally inhibited embryosDominant-negative knockdown of Pax gene functionFunctional rescue of dominant-negative phenotypesDNA binding of Pax gene constructsDevelopmental time-course of Hau-Pax1 protein expression

DiscussionMaternal and zygotic factors in the regulation of embryonic cell lineagePax genes and proteloblast cleavageComparison to other spiralians

AcknowledgmentsSupporting informationReferences